Removing paper and transcription errors from field data collection

I developed a progressive web app to reduce data entry errors while doing five-minute bird counts, an activity traditionally done on paper.

View progressive web app

Client
Personal project
Role
Design and development
Date
2020

Background

I developed this progressive web app to help with the data collection process of my final project at Lincoln University. As part of it, I had to do five-minute bird counts, an activity where you stay stationary for five minutes and record all birds seen and heard. The bird counts follow a standard protocol that is used widely across New Zealand.

Challenge

The traditional data collection process involves carrying paper sheets around, collecting the data and then transcribing it into a digital format. I've done paper counts before and the transcription process is always painful, time consuming and error prone  not to mention having to decipher illegible hand-writing or blotched ink when it rains.

An example of a five-minute bird count done on paper.
An example of a five-minute bird count done on paper

I had a challenging project ahead and wanted to automate the data collection process as much as possible, so decided to create a progressive web app that can be used in areas without cellphone network coverage and that records the counts' data in a digital format.

Process

I was already familiar with the five-minute bird count protocol and knew the exact use case and the setting in which the app would be used. I quickly sketched the UI and created a very basic prototype that guided the development of the app in Vue.js.

Solution

The result is a mobile web app that can be used offline, and where the count data and metadata is stored in a digital format (JSON) that is downloaded after each count.

Tailored to the site

I'd been to the field site before, so had a good idea of what bird species were present there. The main view of the app is populated with a preset list of bird species, but extra species can be added if they are detected on site.

Bird illustrations designed by Fabiola C. Rodriguez Estrada and available for download at the Landcare Research DataStore.

The main view of the app is populated with a preset list of bird species

Adding bird data

Detected birds are added either as being seen or heard, and in this particular project I added distance bands  this can help estimate how many individuals of a species are on a given site. Tapping on a number increments the value for that field; fields can also be edited to correct mistakes or to add very large numbers of birds.

View of the app where information about birds, seen and heard, can be added

Adding metadata

Five-minute bird counts require the collection of additional data. These fields are often present on paper forms as numeric values that are then linked to a key available somewhere else (back of the paper page or footnote). I've exposed the textual descriptions of these fields to make this step more user friendly, but stored the actual values following the standard protocol of using numeric values.

View of the app where the count metadata is added
Data from a five-minute bird count, stored in JSON format.
Data from a five-minute bird count, stored in JSON format

Results

Sixty five-minute bird counts later I had the results stored in a digital format and didn't have to worry about spending hours transcribing the results. Although the development of the app took a while, it was definitely worthwhile as it saved me hours and hours of manual data input and allowed me to focus on getting the project results.

The app being used in the field.
The bird counter app being used in the field